-
Notifications
You must be signed in to change notification settings - Fork 8
/
Copy pathanatomy.yaml
112 lines (112 loc) · 2.65 KB
/
anatomy.yaml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
debug: False
overwrite: False # overwrite outputs of last run if run_id was not changed
ndim: 3
run_id: anatomy
out_dir: output
model_dir: models
log_dir: logs
seed: 42
device: 'cuda:0'
data:
data_dir: prostate158
train_csv: prostate158/train.csv
valid_csv: prostate158/valid.csv
test_csv: prostate158/test.csv
# col names of the image and label files. Pass multiple names as list
image_cols: t2
label_cols: t2_anatomy_reader1
train: True
valid: True
test: False
dataset_type: persistent
cache_dir: /tmp/monai-cache
batch_size: 2
transforms:
# apply multiple transforms to the input. Comment out if a transform should not be applied
# all arguments for specific transformes are passed as keyword arguments directly to the function
prob: 0.175 # independent probability for each transform to be applied
# base transforms
spacing: [0.5, 0.5, 0.5]
orientation: RAS
# imaging specific augmentations
rand_bias_field:
degree: 10
coeff_range: [0.0, 0.01]
rand_gaussian_smooth:
sigma_x: [0.25, 1.5]
sigma_y: [0.25, 1.5]
sigma_z: [0.25, 1.5]
rand_gibbs_nose:
alpha: [0.5, 1]
# affine transformation
rand_affine:
rotate_range: 5
shear_range: 0.5
translate_range: 25
rand_rotate90:
spatial_axes: [0,1]
rand_rotate:
range_x: 0.1
range_y: 0.1
range_z: 0.1
rand_elastic:
sigma_range: [0.5,1.5]
magnitude_range: [0.5, 1.5]
rotate_range: 5
shear_range: 0.5
translate_range: 25
rand_zoom:
min: 0.9
max: 1.1
# resize and crop
# use either `rand_crop_pos_neg_label` or `rand_spatial_crop_samples`
rand_crop_pos_neg_label:
spatial_size: [96, 96, 96]
pos: 2
neg: 1
num_samples: 8
# rand_spatial_crop_samples:
# roi_size: [96, 96, 96]
# num_samples: 8
# intensitiy
gaussian_noise:
mean: 0.1
std: 0.25
shift_intensity:
offsets: 0.2
gaussian_sharpen:
sigma1_x: [0.5, 1.0]
sigma1_y: [0.5, 1.0]
sigma1_z: [0.5, 1.0]
sigma2_x: [0.5, 1.0]
sigma2_y: [0.5, 1.0]
sigma2_z: [0.5, 1.0]
alpha: [10.0, 30.0]
adjust_contrast:
gamma: 2.0
model:
# create a standard UNet with monai (for quick experiments)
# or specify a path to a script with more sophisticated archtitecture
out_channels: 3 # one for each label + background
channels: [16, 32, 64, 128, 256, 512]
strides: [2, 2, 2, 2, 2]
num_res_units: 4
act: PRELU
norm: BATCH
dropout: 0.15
optimizer:
Novograd:
lr: 0.001
weight_decay: 0.01
amsgrad: True
lr_scheduler:
OneCycleLR:
max_lr: 0.001
loss:
DiceCELoss:
include_background: False
softmax: True
to_onehot_y: True
training:
max_epochs: 500
early_stopping_patience: 1000